[lammps-users] Gay-Berne for rods

Craig -

You were correct to be concerned about the singularity. In addition to
the general inverse distance dependence problem, the current distance of
closest approach approximation will result in a singularity under
certain circumstances even when the distance is non-zero. Additionally,
there will be problems with the rotational terms in the potential that
are not a function of distance.

- Mike

Thanks, Mike.

What are the largest aspect ratios one could use and avoid problems?

If I want to make my own "rod" potential, I'm probably better off hacking pair_dipole_cut.cpp and using fix_nve_dipole rather than hacking pair_gayberne.cpp and using fix_nve_asphere, right? In the latter case I need to worry about the full quaternion but for the former there's only the orientation of the dipole vector. The rod potential would just need to set the force and torque for the "dipole" type atoms. Does that sound right?